University of Maryland, Baltimore County
University of Maryland, Baltimore County: UMBC Digital CollectionsNot a member yet
17643 research outputs found
Sort by
A Low Power On-board Processor for a Tongue Assistive Device
No full textIn biomedical wearable devices, patient's convenience and accuracy are the main priorities. To fulfill the patient's convenience requirement, the power consumption, which directly translates to the battery lifetime and size, must be kept as low as possible. Meanwhile, adopted improvements should not impact the accuracy. Therefore, focus on reducing the energy consumption within these devices has already been the subject of a significant amount of research in the past few years. In most wearable devices, all raw data is transmitted to a computer to carry out the required processing. This vast amount of communication leads to a considerable amount of power consumption and the need for a bulky battery, which hinders the device's practicality and patient's convenience. Tongue Drive System (TDS) is a new unobtrusive, wireless, and wearable assistive device that allows for real time tracking of the voluntary tongue motion in the oral space for communication, control, and navigation applications. The intraoral TDS clasps to the upper teeth and resists sensor misplacement. However, the iTDS has more restrictions on its dimensions, limiting the battery size and consequently requiring a considerable reduction in its power consumption to operate over an extended period of two days on a single charge. In this thesis, we propose an ultra low power local processor for the TDS that performs all signals processing on the transmitter side, following the sensors. Implementing the computational engine reduces the data volume that needs to be wirelessly transmitted to a PC or smartphone by a factor of 30x, from 12 kbps to ~400 bps. The proposed design is implemented on an ultra low power IGLOO nano FPGA and is tested on AGLN250 prototype board. According to our post place and route results, implementing the engine on the FPGA significantly drops the required data transmission, while an ASIC implementation in 65 nm CMOS results in 0.128 mW power consumption and occupies a 0.02 〖mm〗^2 footprint. To explore a different architecture, we mapped our proposed TDS processor on the EEHPC many-core. The many-core has a flexible and time saving design procedure. As a result of having a local processor, the power consumption and size of the iTDS will be significantly reduced through the use of a much smaller rechargeable battery. Moreover, the system can operate longer following every recharge, improving the iTDS usability
ANDROID MALWARE DETECTION AND CLASSIFICATION USING MACHINE LEARNING TECHNIQUES
No full textAndroid is popular mobile operating system and there are multiple marketplaces for android applications. Most of these market places allow applications to be signed using self-signed certificates. Due to this practice there exists little or very limited control over the kind of applications that are being distributed. Also advancement of android root kits is making it increasingly easier to repackage existing android applications with malicious code. Conventional signature based techniques fail to detect these malwares. So detection and classification of android malwares is a very difficult problem to solve. We present a method to classify and detect such malwares by performing dynamic analysis of the system call sequences. Here we make use of machine learning techniques to build multiple models using distributions of syscalls as features. Using these models we predict whether given application is malicious or benign. Also we try to classify given application to specific known malware family. We also explore deeplearning methods such as stacked denoising autoencoder (SdA) algorithms and its effectiveness. We experimentally evaluate our methods using a real dataset of 600 malicious applications spread across 38 malware families along with 25 popular benign applications from various areas. We find that deeplearning algorithm (SdA) is most accurate in detecting a malware with lowest false positives while AdaBoost performs better in classifying a malware family
POSITIVE MOOD AND SOCIAL INFORMATION PROCESSING IN PRESCHOOL CHILDREN
No full textPositive mood predicts a broad range of psychological correlates of well-being in children. The present study examined the effects of positive mood on 3 components of social information processing (SIP) in young children including cue encoding, cue interpretation, and response access. The study hypothesized that positive mood and positive emotionality would have main effects on global orientation, response access, and socially competent responding and would also have an interactive effect on these variables. It was also hypothesized that anger would increase anger attribution bias. A computer game, during which participants won or unfairly lost toys, was used to induce happiness and anger in 104 participants recruited from Head Start Centers in urban communities. Participants also were exposed to a neutral stimulus during which they viewed pictures of everyday household and classroom objects. Friedman tests were used to examine the main and interactive effects of positive mood and positive emotionality on cue encoding and response access and the main effect of anger on cue interpretation. Spearman correlations were used to examine the relation between positive emotionality and SIP variables. Results indicated that positive mood has a positive main effect on global orientation and interacts with positive emotionality to predict global orientation such that the effect of positive mood is stronger and more positive for children higher in positive emotionality. Implications for these findings are discussed
Impact of Federally Qualified Health Centers on Rates of Ambulatory Care Sensitive Conditions among Medicaid and Uninsured Populations in Maryland
No full textAmbulatory Care Sensitive Conditions (ACSCs) are conditions for which hospitalization and emergency department (ED) visits can be avoided if a person has better access to ambulatory care services. Rates and costs of hospital admission and ED visits for ACSCs have increased over the decade, especially for people without health insurance and/or on Medicaid. Objectives of this dissertation were to study ACSC rates in Maryland over time, identify areas where ACSC rates had been persistently high, determine factors that were associated with ACSCs, and examine if the expansion of FQHCs had decreased ACSCs in geographical areas over the study period. The study used Maryland hospital discharge data to identify ACSCs based on the Agency for Healthcare Research and Quality (AHRQ) definition of Prevention Quality Indicators (PQIs). A Zip Code Tabulation Area (ZCTA) was used as a unit of analysis. ACSCs and all controlled factors were calculated at the ZCTA level from 2000 to 2010. Negative binomial panel models were used to determine trends, and to estimate the impact of FQHCs and other factors on ACSCs. The study found that ACSC rates among Medicaid and uninsured patients had increased over time for several conditions while such conditions among total populations remained stable or decline. In addition, variations in hospitalization and ED visits for ACSCs existed across Maryland's counties and local areas, but the rates seemed consistent within the areas over time. Proportion of populations living in poverty had the largest and consistently positive relationship with most ACSC hospitalization and ED visit rates. The relationships between ACSCs and other socioeconomic factors are varied by type of condition. Importantly, the expansion of FQHCs had a significant association with lower rates of hospitalization and ED visits for several ACSC conditions. Thus, the expansion of FQHCs is associated with better access to primary care among Medicaid and uninsured populations
EXPLOITING THE ROLE OF POLARITY IN CITATION ANALYSIS
No full textCitation analysis has been widely used as an important factor in the assessment of research impact, which plays a role in advancing an academic field and in recognizing the research reputation and contribution of papers, authors, groups, institutions, and journals. Traditional citation analysis methods are based on the counts of citations or publications. Despite that more recent bibliometric indicators such as the H-index attempts to address the quality of publications, they are still based on statistical counts. Consequently, extant methods for citation analysis have largely overlooked the context in which articles are referenced such as the opinions of the citing authors, and have not been able to resolve the inconsistent findings about the impact of self-citations. The primary objective of this research is to address the limitations of the state-of-the-art citation analysis methods by examining the context as well as content of citations. The context is manifested in the polarity and the location of citations. Specifically, the polarity of citations is classified through content analysis into one of the three categories: positive, negative and neutral. The location of citations was measured in terms of common structural components of a research article. The secondary objective of the research is to examine the possible influence of self-citation on the outcome of citation analysis. The proposed methods for citation analysis are evaluated using real-world data collected from digital databases of publications. Using the traditional count based method as the baseline, the experiment results demonstrate that the proposed methods not only offer a new perspective for citation analysis but also provide an alternative explanation for the influence of self-citations
Using Machine Learning Techniques to Identify Source Code Documents
No full textThis thesis examines the application of document classification techniques to collections of source code for the purpose of semantic annotation. Typical approaches to this problem in previous research have involved static analysis, which requires an understanding of some program semantics already, dynamic analysis, which requires program execution, or learning from metadata associated with a program. The methodology of this work is unique in that it utilizes machine learning methods to identify the behavior of algorithms based solely on syntactic features of the source code. Applications include indexing source code repositories to expedite searching therein, and recognizing inconspicuous programming faults that can lead to software producing undesired results. The methodology begins with the extraction of features from source code. The performance of both decision trees and support vector machines is analyzed. In addition, a complementary approach is evaluated that uses unsupervised clustering to find natural groupings and concept descriptions in a set of source code fragments. We provide an analysis of the effects of varying n-gram degrees, feature expressiveness and feature selection on the accuracy of the classifiers, and the structure and concepts generated by the clustering approach. Even with minimally expressive features, both the support vector machine and decision tree models are able to categorize source code samples based on the algorithm they implement with high accuracy. Most classification errors tend to occur between semantically similar algorithms, but disparate algorithms are often distinguishable
USERS' PRIVACY AND SECURITY BEHAVIORS ON MOBILE DEVICES.
No full textPreferences and behaviors for privacy management with mobile applications are difficult to capture. Previous measures are mostly based on self-report data, which often does not accurately predict actual user behavior. A deeper understanding was sought, gleaned from observing actual practices. This thesis analyzes 11,777 applications from the Google Play marketplace in order to determine the impact of privacy settings on purchase behavior. This was done by looking at the effect of the number of privacy concessions as well as the effect of individual concessions and category on number of downloads. It was found that users of paid applications do not have a preference for fewer privacy concessions. This study further reinforces the disconnect between the user's often stated preference for privacy and their actual behavior – a discrepancy known as the “privacy paradox ”. Theoretical and practical implications are discussed
INVESTIGATING THE CHLAMYDOMONAS REINHARDTII CALVIN CYCLE ENZYME GENES FBP1 AND SBP1: REGULATION AND EFFECT OF OVEREXPRESSION ON GROWTH
No full textMicroalgae have great potential for generating biofuels, pharmaceuticals, and other commercially valuable products. The goal of this thesis project was to improve this potential by investigating the expression and function of two key Calvin cycle enzymes in Chlamydomonas reinhardtii. Overexpression of sedoheptulose-1,7-bisphosphatase (SBPase) and fructose-1,6-bisphosphatase (FBPase) improves carbon fixation and growth rate in higher plants, but nothing has been reported regarding the role of these enzymes in photosynthetic output and biomass in C. reinhardtii, a model green alga. I first set out to determine how overexpression of C. reinhardtii SBPase affects Chlamydomonas growth by making nuclear transgenic lines expressing the SBPase-encoding gene SBP1 under control of the HSP70A/RBCS2 promoter and RBCS2 3'UTR. One transgenic line that overexpressed SBPase showed a promising result by reaching the exponential and stationary growth phases faster than the control strain, but expression of transgenic SBPase in this strain diminished significantly within 6 months, so I next set out to overexpress SBP1 and the FBPase encoding gene (FBP1) directly in the chloroplast, to avoid the possibility of gene silencing. To this end, I generated new chloroplast expression vectors with C. reinhardtii FBP1 and SBP1 coding sequences synthesized with C. reinhardtii chloroplast codon-bias and flanked with 5'psbD and 3'psbA regulatory sequences. The FBP1 vector integrated properly into the chloroplast genome and all FBP1 transformants tested accumulated ~4-fold increased levels of FBPase protein and 1.5-fold increased levels of FBPase activity; unfortunately, however, none of the SBP1 transformants expressed SBPase protein. Surprisingly, the FBPase-overexpressing transformants did not grow faster than the wild type, and in fact under mixotrophic conditions with 5% CO2 they grew more slowly and reached a maximum cell density that was 1.5-fold lower than the wild type did, and under photoautotrophic conditions they grew more slowly and reached maximum cell densities that were 1.4-fold lower (atmospheric CO2) and 1.7-fold lower (5% CO2) than the wild type. To learn more about regulation of FBP1 and SBP1 mRNA and FBPase and SBPase protein levels I cultured C. reinhardtii cells photoautotrophically under 12 hour light:12 hour dark conditions and analyzed mRNA levels by RT-qPCR and protein by western blot. I found FBP1 and SBP1 transcripts to be highly inducible by light, up to ~8 and ~9 fold, respectively, while FBPase and SBPase protein levels were only modestly elevated in the light ~2.5 and ~6 fold, respectively. So in conclusion, I determined that Chlamydomonas FBP1 and SBP1 appear to be light-regulated genes, that overexpressing SBPase might be a promising strategy for improving microalgal growth, but that overexpression of FBPase has a neutral or detrimental effect on growth and biomass productivity in C. reinhardtii, depending on growth conditions
Identifying Malware Using N-Gram Clustering Metrics
No full textWe identify a new method for detecting malware within a network that can be processed in linear time. In the digital age, more files are transferred between individuals and systems that have the potential to contain malignant processes. Traditional malware detection and analysis is performed by signature based operations or by hashing current files. A malicious attacker can quickly change found signatures or change various processes to defeat hash based detection. We need a way to quickly identify malicious files to stage them for quarantine and further analysis. In this thesis we observe the previous methods used to detect malware and develop a new process to identify malware using n-gram analysis to cluster malware specimens by their similarity to each other. Specimens from a well-known malware family are used in this demonstration
The Visual Ecology of Stomatopod Larvae
No full textDespite researchers' interest in adult stomatopod vision, only a handful of studies have investigated stomatopod larval visual ecology, the function of the larval compound eye, or its transition into the adult eye structures. In order to address this knowledge deficit, I tested several hypotheses surrounding larval eye physiology and ontogeny using microspectrophotometry (MSP) and electroretinography (ERG). MSP was used to investigate visual pigment diversity in the retinas of different species of stomatopod larvae. Together with data previously published, these data provide further support that larvae possess a single spectral class of photoreceptor. Surprisingly, the peak spectral absorbance varied significantly among sympatric species, with four species falling into a short-wavelength and the remainder into a long-wavelength sensitivity class. These results may be explained by differences in behavioral ecology or by different developmental stages of individual test subjects. To investigate ontogenetic changes in retinal function, ERG intensity responses were measured from different stage retinas of a single species, Squilla empusa. Initial MSP data from transitioning retinas found no evidence of visual pigment absorption in newly emerged adult retinas, implying a putative delay in photoreceptive function. Robust ERG responses to light stimuli, however, were observed from the earliest presence of emerging adult retinas, rejecting this hypothesis. These data also suggest an increase in the response dynamics of stomatopod retinas with ontogeny as well as an increase in irradiance sensitivity during the double-retina phase. To investigate the visual ecology of stomatopod larval light-reflecting structures, I characterized the function and structure of eyeshine overlying the retina. In situ photography of larvae eyeshine demonstrated its function as a contrast reducer, or camouflage. Calculations of eyeshine reflectance spectra in their natural setting revealed novel, spectral matching with the background environment. Eyeshine structures were characterized via transmission electron microscopy (TEM) and identified as amorphous, three-dimensional photonic crystal arrays of spherical vesicles that coherently scatter light. The photonic mechanism of eyeshine production was tested using a modified Bragg-theory model and dimensions from TEM micrographs. These eyeshine data as well as data regarding the physiology of stomatopod larval vision provide a strong foundation for future investigations into the ontogeny of vision in stomatopod crustaceans